Dynamic Article LinksC<Journal ofMaterials Chemistry

Cite this: J. Mater. Chem., 2011, 21, 7401

www.rsc.org/materials PAPER

Publ

ishe

d on

11

Apr

il 20

11. D

ownl

oade

d by

Car

negi

e M

ello

n U

nive

rsity

on

22/1

0/20

14 1

8:07

:12.

View Article Online / Journal Homepage / Table of Contents for this issue

Preparation of porous composite ion-exchange membranes for desalinationapplication†

Chalida Klaysom,a Roland Marschall,a Seung-Hyeon Moon,b Bradley P. Ladewig,ac G. Q. Max Lu*a

and Lianzhou Wang*a

Received 29th November 2010, Accepted 15th March 2011

DOI: 10.1039/c0jm04142d

Via a two-step phase inversion technique, composite membranes with controllable porosity and

a significant improvement of electrochemical properties were successfully prepared. The presence of

surface functionalized mesoporous silica (SS) as inorganic fillers in the sulfonated polyethersulfone

(sPES) polymer matrix was proved to have a great impact on the resultant membrane structure, which

subsequently led to significantly enhanced ionic conductivity of the membranes. The correlation among

inorganic fillers, composite structures, electrochemical properties and desalination performance by

electrodialysis (ED) was discussed in detail. The optimal membrane was the composite with 0.2 wt% SS

loading, which possessed a good ionic conductivity of 5.554 mS cm�1, a high selectivity with 0.95

transport number while maintaining good mechanical strength and thermal stability. Moreover, the

performance of this membrane in ED was comparable to a commercial membrane (FKE), exhibiting

a current efficiency of 0.84 and 3.82 kW h kg�1 of salt removed.

Introduction

Ion-exchange membranes have attracted great attention from

both academic and industrial fields, due to their versatile

potential applications including water purification and food

processing.1–10 Many attempts have been made by a number of

research groups to develop robust membranes with desirable

properties for different applications. These include (1) a search

for new polymers with highly modified chemical structures

enhancing membrane conductivity and electrochemical proper-

ties, (2) surface and structure modification of existing commer-

cial membranes, and (3) improvement of considerably cheaper

polymer in the market by blending other polymers or incorpo-

rating some additives into the parent matrix.11–13 Though the

concept of combining two distinct materials forming a new

composite that keeps desirable properties of both components

has been well established, it has not been applied much in the

development of ion-exchange membranes especially in desali-

nation applications. In this work we aim to design a new type of

aARC Centre of Excellence for Functional Nanomaterials, School ofChemical Engineering and Australian Institute of Bioengineering andNanotechnology, The University of Queensland, Qld, 4072, Australia.E-mail: l.wang@uq.edu.au; Fax: +61 7 3365 4199; Tel: +61 7 3365 4218bGwangju Institute of Science and Technology, School of EnvironmentalScience and Engineering, Gwangju, 500-712, Republic of KoreacMonash University, Department of Chemical Engineering, Vic, 3800,Australia

† Electronic supplementary information (ESI) available: Supplementaryinformation. See DOI: 10.1039/c0jm04142d

This journal is ª The Royal Society of Chemistry 2011

hybrid ion-exchange membranes by introducing inorganic fillers

to the sulfonated polyethersulfone (sPES) polymer matrix.

Sulfonated aromatic hydrocarbons such as sPES have been

extensively used as an alternative electrodriving membrane due

to their low cost, excellent mechanical, thermal and chemical

stabilities. Moreover, this type of polymers is also easy to process

and has been widely used as the polymer matrix alternative in

replacing Nafion for composite membranes in fuel cell applica-

tion.10,14–16 A variety of inorganic fillers such as sulfonated

polyhedral oligomeric silsesquioxane (sPOSS), phosphotungstic

acid, boron phosphate, clay, and silica were introduced into the

sPES to prepare a composite membrane by either sol–gel or

blending method.17–21 A few successful examples of composite

proton-exchange membranes with improved ion-exchange

capacity and conductivity suitable for fuel cell application have

been reported.17–27 For instance, Choi et al. recently reported the

preparation of sulfonated polysulfone/sPOSS fiber composite

membranes with superior ion-exchange capacity and high proton

conductivity even at very low relative humidity.17 Note that this

type of membranes required several steps to prepare, which may

be disadvantageous for scale-up preparation. Despite the success

in composite proton exchange membrane preparation, the

development of composite membranes for desalination applica-

tion has been very rare. In our previous work, composite

membranes containing a small amount of surface functionalized

mesoporous silica (SS) in the sPES polymer matrix were

prepared by a simple procedure. The resultant composite

membranes exhibited enhanced electrochemical properties while

still retaining excellent mechanical strength and thermal

J. Mater. Chem., 2011, 21, 7401–7409 | 7401

Publ

ishe

d on

11

Apr

il 20

11. D

ownl

oade

d by

Car

negi

e M

ello

n U

nive

rsity

on

22/1

0/20

14 1

8:07

:12.

View Article Online

stability.28 However, due to the dense structure of the obtained

membranes prepared by the widely used solvent evaporation

method, their ionic conductivity was still very low, being

incomparable with commercial membranes on the market.

In the present work, we report an innovative structural

modification pathway, namely the two-step phase inversion

technique, to tailor the membrane porosity of the sPES–SS

composite membranes, which in turn led to significantly

improved ionic conductivity for the membranes. The relationship

among porosity/structure, physicochemical and electrochemical

properties of the resultant membranes was investigated in detail.

Encouragingly, the performance of this new type of membrane in

ED desalination was comparable to that of a commercial FKE

membrane.

Experimental

Material synthesis

The synthesis procedure and material characterization of sPES

and SO3H-functionalized mesoporous SiO2 (SS) were described

in more detail elsewhere.13,28 The ion-exchange capacity of SS

was measured by back titration. The structure and morphology

of the SS were studied by transmission electron microscopy

(TEM, JEOL 1010) and X-ray diffraction (XRD, Rigaku Min-

iflex) under copper radiation (CuKa, 30 kV and 15 mA).

Membrane preparation

Composite membranes were prepared by a two-step phase

inversion procedure. In the first step, a polymer solution con-

sisting of 25 wt% sPES in dimethylformamide (DMF, Sigma)

was prepared, and then various amounts of SS (0–2 wt%) were

added into the polymer solutions at 60 �C for 4 h under stirring.

Sonication was applied to the mixture solution before it was cast

on glass substrates using a doctor blade with a casting thickness

of ca. 0.40 mm. In the second step, the cast film was dried in

a vacuum oven at 60 �C for 10 min and then precipitated in a 60–

70 �CDI water bath. The formed membrane sheet was peeled off

from the glass substrate and kept in DI water. A series of

membranes were named by considering the wt% of SS added into

the polymer matrix. For instance, 0.2SS represents the composite

membrane of sPES containing 0.2 wt% of SS. The prepared

membranes were then treated in hot water for 2 h and then in

1 mol dm�3 HCl for 24 h. Subsequently, the membranes were

rinsed with DI water and kept in 1 mol dm�3 NaCl solution. All

membranes were equilibrated in working solution for at least 6 h

before use.

Membrane characterization

Membrane morphology. Scanning electron microscopy (SEM,

JEOL 6300) was used to observe the morphology and structure

of the prepared membranes. To obtain a sharp cross-sectional

surface the samples were fractured in liquid nitrogen, and then

the captured water was dried out overnight in a freeze dryer to

preserve their structure.

Ion-exchange capacity (IEC) and water uptake of ion-exchange

membranes. Ion-exchange capacities were measured by a back

7402 | J. Mater. Chem., 2011, 21, 7401–7409

titration technique. Firstly, the cation-exchange membrane was

soaked in 1 mol dm�3 HCl for at least 6 h. After that, the

membrane was washed with DI water to remove the excess HCl

and then was immersed into 1 mol dm�3 NaCl solution for

another 6 h. The number of displaced protons from the

membrane was determined by titration with 0.01 mol dm�3

standard NaOH solution using phenolphthalein as an indicator.

Then the membrane was soaked in DI water for 24 h or more.

Afterward, the membrane was taken out and the excess water on

the surface was removed with tissue paper. The wet weight of

membrane was measured. Then the membrane was dried at 50 �Cin an oven for 10 h or until there were no weight changes in the

membranes. The dry weight of the membrane was recorded. The

IEC and water uptake of membranes were then calculated using

eqn (1) and (2):

IEC ¼ ab

Wdry

(1)

Water uptake ¼ Wwet �Wdry

Wdry

(2)

where a is the concentration of NaOH solution (mol dm�3), b is

the volume of NaOH solution used (dm3), Wdry is the dry

weight of the membrane and Wwet is the wet weight of the

membrane.

Electrochemical properties of ion-exchange membranes

Membrane resistance. The resistance of membranes was

measured by impedance spectroscopy (IS) using a Solarton 225B.

The working ion-exchange membrane was equilibrated in

0.5 mol dm�3 NaCl before being placed in a two-compartment

cell between platinum electrodes with an effective area of 1 cm2.

The test was carried out in a frequency range of 1–106 Hz with an

oscillating voltage of 100 mV.29 The resistance corresponding to

the phase angle closest to zero in the Bode diagram was recorded.

Then, the membrane resistance (Rmem) was calculated by

subtraction of the electrolyte resistance (Rsol) from the

membrane resistance equilibrated in the electrolyte solution

(Rcell), according to the equation Rmem ¼ Rcell � Rsol. The

conductivity of membranes (s, S cm�1) was then calculated by

eqn (3):\displaystyle

s ¼ L

RmemA(3)

where Rmem was the resistance of membranes, L is the thickness

of membranes (cm), and A is the effective area of the membranes

(cm2).

Membrane potential and transport number. The membrane

potential was measured with dilute solutions in order to render

negligible the activity coefficient of the electrolyte.30 The

investigated membrane was placed in a two-compartment cell,

separating two NaCl concentration solutions of 0.01 mol dm�3

and 0.05 mol dm�3, respectively. The potential difference (Em)

across the membrane was measured at room temperature using

a multimeter which was connected to two Ag/AgCl reference

electrodes. The transport number, �ti, was calculated by the

following equation:4

This journal is ª The Royal Society of Chemistry 2011

Publ

ishe

d on

11

Apr

il 20

11. D

ownl

oade

d by

Car

negi

e M

ello

n U

nive

rsity

on

22/1

0/20

14 1

8:07

:12.

View Article Online

Em ¼ RT

F

�2ti � 1

�ln

�C1

C2

�i

(4)

where R is the gas constant, F is the Faraday constant, T is the

absolute temperature, C1 and C2 are the concentrations of elec-

trolyte solutions in the testing cell, respectively.

Fig. 1 ED cell configuration (a) and schematic diagram of the test in ED

system (b).

Current–voltage (i–v) characteristic and chronopotentiometry.

The i–v curve and chronopotentiogram of the membranes in

0.025 mol dm�3 NaCl were measured at room temperature using

the same two-compartment cell. The cell was connected to

a Solartron Multistat 1480 by two Pt electrodes. For i–v curves,

a stepwise potential difference was applied across the membrane

and the corresponding current was measured. The chro-

nopotentiogram was carried out at a constant applied current

density of 3 mA cm�2 and the corresponding potential was

recorded every 1 s for a period of 300 s automatically.

Thermal and mechanical stabilities of membranes

Thermal stabilities. Thermogravimetric analysis (TGA)

(Mettler Teledo) was used to study the thermal stability of the

membranes in the temperature range of 25–800 �C with a heating

rate of 10 �C min�1 under a nitrogen flow of 20 cm3 min�1.

Mechanical properties. The mechanical properties of wet

membranes were measured by the tensile test at room tempera-

ture using an Instron 5800 at a speed of 2 mm min�1. The tested

samples were equilibrated in DI water for 24 h and then cut into

a rectangular shape with dimensions of 50 mm � 5 mm. The

gauge length of each specimen was 14 mm. At least five speci-

mens from each sample were tested.

Desalination by electrodialysis

The performance of the prepared membranes was tested in

a custom designed ED cell. The cell consists of 5-compart-

ment chamber made of Perspex sheet. Each chamber provides

4 cm2 active area for membranes with an o-ring to prevent

leakage during the testing. The scheme of the ED setup and

the membrane configuration in the cell are illustrated in

Fig. 1.

The ED of NaCl solution was carried out with potentiostatic

modes. The feed solution (150 cm3 of 0.2 mol dm�3 NaCl) was

circulated at 30 cm3 min�1 through the dilute and concentrated

compartments. The concentrated compartments were adjacent to

the electrode rinse compartments in which 150 cm�3 of 3 wt%

Na2SO4 was circulated at 30 cm3 min�1. The commercial cation-

exchange membranes FKE were placed at the positions C1 and

C3 in Fig. 1(a) next to the electrode to separate the electrode

solution from the product solution. The composite membrane as

a cation-exchange membrane was placed in position C2 and

commercial anion-exchange membrane was placed in position A

to complete the cell. The performances of the prepared

membranes were compared with the commercial membrane in

terms of flux, current efficiency and energy consumption calcu-

lated by the following equations.

Flux ¼ DN

At(5)

This journal is ª The Royal Society of Chemistry 2011

h ¼ FzDN

ncÐIdt

(6)

P ¼ FzU

3600hMNacl

(7)

where h ¼ current efficiency of the dilute, F ¼ Faraday constant

(96 487 As mol�1), DN ¼ Cdn�1Vd

n�1 � CdnVd

n (mole), Cd ¼concentration of dilute (mol dm�3), Vd ¼ volume of the dilute

(dm�3), P ¼ power consumption of NaCl (kW h kg�1), nc ¼number of cell pair, I ¼ current (A), U ¼ applied voltage (V),

MNaCl ¼ molecular weight of NaCl (58.45 g mol�1), A ¼ active

surface area (cm2), and t ¼ time (s).

The ion concentration from each compartment was deter-

mined by a conductometer and the pH of solution from each

chamber was measured to indicate the water electrolysis that may

occur during the process. The volume change in each reservoir

was recorded at the start and end of the experiment.

Results and discussion

Structure influence of the inorganic fillers on composite

properties

Following the same procedure in our previous work,22 the SS

with a particle size of 100–150 nm was firstly synthesized. The

highly ordered mesoporous materials were evidenced by TEM

images (Fig. 2(b and c)). Moreover, the small angle XRD of as-

synthesized sulfonated SiO2 showed well resolved reflections of

2D-hexagonal mesostructure with indexes of (100), (110), (200),

and (210), confirming the highly ordered pore structure of the

particles. The ion-exchange capacity of the SS particle was

J. Mater. Chem., 2011, 21, 7401–7409 | 7403

Fig. 2 Small angle XRD patterns (a) and TEM (b and c) images of sulfonated mesoporous silica (SS) at different magnifications.

Publ

ishe

d on

11

Apr

il 20

11. D

ownl

oade

d by

Car

negi

e M

ello

n U

nive

rsity

on

22/1

0/20

14 1

8:07

:12.

View Article Online

estimated to be 1.8 mequiv g�1 by back titration, which is

consistent with those reported in previous works.13,28

The SS particles were first mixed with polymer solution of

sPES and DMF, followed by the two-step phase inversion

process to form membrane sheets.31 FTIR spectrum was used to

investigate the chemical structure of each component and

possible intermolecular interaction. As shown in Fig. S1†, IR

vibration bands at around 1140 and 1205 cm�1, characteristics of

SO3H groups attached to the silica, and at 1030 cm�1 for sulfo-

nate groups of sPES were observed. These indicate the presence

of ion-exchangeable groups in both sPES and SS components.

However, the FTIR bands of the SS could not be clearly iden-

tified in the composite, possibly due to the very small proportion

of SS in the composite or band position overlapping of the

components.32

The morphology of the obtained composites was shown to be

highly porous (Fig. 3). Compared to the pristine membrane, the

presence of additives increased the porosity of the membranes.

Not only the porosity of the membranes increased with SS

loading, but also the pore size of the membrane. At higher

loading the SS particle tends to form clusters, thus resulting in

a bigger pore size. Though the SS cluster formation became more

pronounced at high percentage of SS loading, these primary

Fig. 3 SEM images of (a1 and 2) sPESmembranes, (b1 and 2) composites wit

wt% SS.

7404 | J. Mater. Chem., 2011, 21, 7401–7409

clusters were still well distributed throughout the entire structure

of the polymer matrix.

The key properties of ion-exchange membranes including

water uptake, IEC, conductivity and transport number were

summarized in Table 1. As mentioned earlier, incorporating high

surface functionalized silica was expected to improve the IEC

and water uptake and thus enhances the conductivity and

transport properties of the membranes. The water uptake of the

membrane was significantly improved due to three main effects

of the additives; the additional hydrophilic functional groups

carried by the high surface area SS, the inherit property of silica

to keep both physical and chemically bound moisture, and pore

formation of the membrane contributed by SS, creating free

space to absorb more water.33 Since water channel promotes the

migration of the ionic species, the conductivity of the membrane

increased, following the same trend of the water uptake. Like-

wise, the IEC of the membrane increased with increasing

the loading amount of SS and peaked at low percentage (around

0.2 wt%) of SS. The decrease of IEC when more SS was added

may be due to the interaction among functional groups of the

inorganic filler clusters, because agglomeration of fillers can

cause the loss of the accessibility of the functional groups in SS

and thus the decrease of IEC, resulting in lower water content. In

h 0.2 wt% SS, (c1 and 2) 0.5 wt% SS, (d1 and 2) 1 wt% SS, and (e1 and 2) 2

This journal is ª The Royal Society of Chemistry 2011

Table 1 Physicochemical and electrochemical properties of the prepared membranes compared with commercial membranes

Name Water uptake (%) IEC/mequiv g�1 s/mS cm�1 Transport number Free volume fraction

0SS 38.67 0.75 0.336 0.90 0.4330.2SS 92.00 0.83 5.554 0.95 0.740.5SS 85.00 0.78 1.460 0.95 0.701.0SS 79.00 0.73 1.330 0.96 0.692.0SS 55.00 0.69 0.960 0.91 0.67FKEa 48.10 1.20 3.834 0.95 NA

a A commercial cation exchange membrane from FumaTech, Germany. The FKE is made of sulfonated poly(ether ether ketone). The wet thickness ofthe commercial membrane was 50 � 10 mm which is similar to those prepared membranes.

Publ

ishe

d on

11

Apr

il 20

11. D

ownl

oade

d by

Car

negi

e M

ello

n U

nive

rsity

on

22/1

0/20

14 1

8:07

:12.

View Article Online

addition, more inorganic agglomerate (>1wt% SS) in the

composite means reduced proportion of sPES in the membrane.

Thus the ion-exchangeable functional groups contributed from

sPES could also be reduced. For these reasons, the increased

inorganic fillers and pore cavity did not lead to better membrane

performance. The pore volume or free volume fraction in the

composite membranes again suggests the optimal condition of

0.2 wt% SS, calculated by the following equation:

Free volume fraction ¼ rwet � rdry

rw(8)

where rwet, rdry and rw are the density of wet membrane, dry

membrane and water at room temperature, respectively.

The results of morphology and properties as described earlier

revealed that incorporating inorganic fillers had a significant

impact on the membrane structure which was also directly

related to the properties of the membranes. The influence of these

inorganic fillers was schematically described in Fig. 4. In the

phase inversion process for membrane formation, the final

Fig. 4 Schematic diagram of pore formation with the presence of inorganic

functional groups.

This journal is ª The Royal Society of Chemistry 2011

structure of the membrane is determined largely by the property

of the casting polymer solution and its viscosity.34–37 The less

viscous the polymer solution is, the faster the solvent/nonsolvent

exchange rate is, resulting in a more porous membrane structure.

The pore formation of the composite in the presence of SS may

be attributed to the incompatibility between the inorganic fillers

and the organic polymers, creating a polymer-poor phase. As

a result, the high exchange rate of solvent/nonsolvent around the

inorganic fillers creates pore cavity capturing the SS particles.

When more SS was added in the polymer matrix, the particles

tend to form bigger clusters and thus bigger pores were

produced. It can be seen from the SEM image, at more than 1 wt

% SS loading, the obvious large cavities with pore interconnec-

tion were obtained. When the inorganic fillers formed large

clusters, their surface functional groups were reduced and also

started to interfere with the charged functional groups of the

polymer matrix. As a consequence, the physicochemical and

electrochemical properties of the composite declined after

reaching a maximum value.

fillers and the charge interfering between inorganic clusters and polymer

J. Mater. Chem., 2011, 21, 7401–7409 | 7405

Publ

ishe

d on

11

Apr

il 20

11. D

ownl

oade

d by

Car

negi

e M

ello

n U

nive

rsity

on

22/1

0/20

14 1

8:07

:12.

View Article Online

The thermal and mechanical properties of these membranes

were investigated by TGA and tensile test, respectively (Fig. 5).

Three-step weight loss was observed for the membranes. The first

weight loss at the temperature below 100 �C is associated with the

release of the absorbed moisture. Then, the slight loss at around

280 �C can be assigned to the decomposition of sulfonate

groups,38 and finally dramatic loss occurring at around 500 �C is

due to the degradation of polymer chains. From the TGA curve,

it is observed that the membranes exhibited a better thermal

stability in the presence of inorganic fillers. This may be attrib-

uted to the interaction among the functional groups.

The mechanical testing was carried out under wet conditions

and the results are presented in terms of Young’s modulus,

tensile stress, and strain as shown in Fig. 5(b and c). Incorpo-

rating inorganic fillers enhanced the energy distribution

throughout the membranes, thus the membranes can withstand

more pressure. However, in the presence of additives the

membranes became more elastic and more elongated under

the tensile force. As a consequence, the Young’s modulus of the

membranes decreased. At high degrees of loading, membranes

became more brittle, resulting in the increment of the Young’s

modulus.

For comparison, the thermal and mechanical properties of the

commercial FKE membrane, which is made of sulfonated poly

(ether ether ketone), were acquired from the supplier. It was

informed that the FKE possesses high thermal stability above

270 �C with a Young’s modulus of 1124 MPa, a tensile stress of

36.39 MPa and a tensile strain up to 195.26% at 95% RH.

Compared to the FKE membrane, our prepared membranes

showed comparable thermal stability but lower mechanical

stability. However, one should keep in mind that the testing

conditions from the company and our experiment were different.

The mechanical properties of the commercial membrane were

tested under 95% RH, while our membrane was tested under the

fully hydrated condition. Nevertheless, it was clear to conclude

from the results of this section that our composite membranes

exhibited excellent thermal stability up to 280 �C and decent

mechanical strength up to 600 MPa, which are sufficient for

application in desalination.

Fig. 6 Chronopotentiograms (a) and their derivative potential vs. time

curve (b) of the synthesized membranes compared with commercial

membranes tested at room temperature in 0.025 mol dm�3 NaCl.

Electrochemical behavior of the composite membranes

It is of crucial importance to understand the electrochemical

behavior of the membranes and to predict their performance and

behavior in a real application. The electrochemical behavior of

Fig. 5 Thermal (a) and mechanical properties (b: Young’s m

7406 | J. Mater. Chem., 2011, 21, 7401–7409

as-synthesized membranes was studied utilizing well known

chronopotentiometry and current–potential (i–v) techniques.

Chronopotentiometry is a powerful tool for studying transport

phenomena and concentration polarization near membrane

interfaces. In this work, it was used for investigating the homo-

geneity of membrane surface as the ion-exchange membranes are

generally considered inhomogeneous in the micro-scale consist-

ing of conducting and non-conducting regions. By applying

a constant current density to the system and measuring the

corresponding potential as a function of time, the typical chro-

nopotentiograms with three-stage potential differences were

obtained and shown in Fig. 6(a). After the constant current

density is applied to the system, the concentration polarization

arises, resulting in the concentration gradient and slow increase

of potential at the first stage. When the concentration near the

membrane interface of the depleting solution decreases to nearly

zero, the potential rapidly increases before reaching the steady

state at the end. The time at which the potential transition occurs

is called the transition time (s). The transition time can also be

expressed by the well known Sand’s equation:

s ¼ ðC0ziFÞ2pD4i2ðti � tiÞ2

(9)

where i is the current density, C0 is the concentration of elec-

trolyte, and zi is the valence of the ith ion. It is worth noting that

the eqn (9) was proposed under the assumption of entire con-

ducting surface of the membranes. The s deprived from Fig. 6(b)

odules, c: strain and stress) of the prepared membranes.

This journal is ª The Royal Society of Chemistry 2011

Publ

ishe

d on

11

Apr

il 20

11. D

ownl

oade

d by

Car

negi

e M

ello

n U

nive

rsity

on

22/1

0/20

14 1

8:07

:12.

View Article Online

of the derivative dE/dt at the maximum value where the rapid

potential increase occurred is listed in Table 2 and compared with

the ideal value estimated from eqn (9). It clearly showed the

deviation between those values obtained experimentally and

theoretically. This can be explained by the effects of surface

inhomogeniety of the membranes. Considered ion-exchange

membranes as microheterogeneous membranes with reduced

conducting surface, the local current density at the conducting

regions is higher than the overall current density applied to the

system. As a result, the time required to deplete the ionic species

became shorter. The current density in the conducting region (i*)

was expressed as a function of overall current density or so-called

superficial current density (i) and the fraction of the conducting

region (3) as shown below;

i* ¼ i/3 (10)

And thus the 3 can be estimated by the derived Sand’s equation

given as:

3 ¼ 2is1=2ðti � tiÞC0ziFðpDÞ1=2

(11)

The fraction of conducting surface, 3, including all characteristic

values from chronopotentiograms is summarized in Table 2. The

starting potential (E0) and the potential at the steady state (Emax)

varied in each membrane. These characteristic values are

believed to be dependent on the resistance of membranes and the

testing systems. The fraction of conducting region increased with

the amount of SS added and peaked at 0.2 wt% loading which is

in a good agreement with the result in the previous section. The

results clearly showed that the shape of chronopotentiogram and

their characteristic values are highly influenced by the surface

heterogeneity of the membranes. While the membrane with more

surface heterogeneity (0SS) exhibits more diffused curve with

more interval time (Dt) where the derivative dE/dt is not zero as

can be seen in Fig. 6(b), the more surface homogeneous

membranes with the presence of SS shows more defined transi-

tion time with smaller interval time.

The i–v characteristic was also used to investigate the

concentration polarization of membrane interface and the

limiting operating point so-called the limiting current density

(LCD). By recording the corresponding current density with the

stepwise potential difference, typical i–v curves of the prepared

membranes were obtained as shown in Fig. 7. The linear rela-

tionship of current and potential in the beginning stage followed

the Ohm’s law. Subsequently, a plateau was reached when the

concentration of the dilute solution near the membrane interface

Table 2 Characteristic values from chronopotentiograms

Sample �ti sa/s sb/s 3 E0/V Emax/V Dt/s

0SS 0.90 32.2 24 0 0.864 3.45 3.94 260.2SS 0.95 26.6 26.5 0.997 3.14 3.83 200.5SS 0.95 26.6 26.5 0.997 3.14 3.90 241SS 0.96 25.7 23.0 0.946 3.20 3.93 242SS 0.91 30.9 23.0 0.862 3.25 3.95 19FKE 0.95 26.6 26.0 0.988 3.10 3.94 24

a Estimated from eqn (9). b Deprived experimentally fromchronopotentiogram.

This journal is ª The Royal Society of Chemistry 2011

dropped nearly to zero due to the concentration gradient that

arose from concentration polarization. The LCD can then be

estimated from intersection of the tangents from these two

regions. The corresponding current above the plateau region can

be referred to a ‘‘so-called’’ over-limiting region which was

believed to be influenced dominantly by the convection effects.

The derivative dE/di is deprived from the original i–v curve and

plotted as a function of i and E as shown in Fig. 7(b and c),

respectively, in order to avoid the ambiguous estimation of the

LCD from the ill-defined characteristic curve and to better

visualize the plateau length. According to the concentration

polarization theory, the LCD was expressed in terms of diffusion

boundary layer thickness (d), diffusion coefficient (D), and

transport number as shown in eqn (12):

ilim ¼ jzjCFDd�ti � ti

� (12)

It is clear from the eqn (12) that under the same testing condi-

tions, the LCD depends on the transport number of membrane.

Note that the dwas considered constant in this work as this value

is a function of hydrodynamic property such as feed flow or

stirring rate which was also controlled in the experiment. The

lower the transport number the membrane has, the higher the

LCD of the membrane was expected.

However, the result showed that our composite membranes

with higher transport number possessed the higher LCD. This

may be attributed from the reduced conducting area of the

microheterogeneous membranes explained earlier. The superfi-

cial LCD (i) deprived from Fig. 7(b) and the local LCD (i*) of the

conducting region estimated from eqn (10) were compared in

Table 3. The results showed that the local LCD (i*) is propor-

tional to the reciprocal of the transport number of the

membranes, theoretically expressed in eqn (12). It is also inter-

esting to take a note of the plateau length of the prepared

membranes that decreased with the increase of % loading and the

fraction of the conducting region. The more heterogeneous the

membrane surface is, the longer the plateau length is. This is

consistent with reports in the literature.4,30,39–42 However, the

plateau length of composites with 2 wt% SS was much smaller

than the parent membrane even though they possessed close 3

values and similar properties. This may be explained by the effect

of geometrical heterogeneity reported by early studies,39,40

revealing that not only the fraction value (3) but also the

arrangement and distribution between conducting and non-

conducting regions needed to be taken into account for

explaining the plateau shortening or prolonging behavior. If the

space distance between the two compartments is in the same

magnitude with the boundary-layer thickness, the shortening

plateau length of heterogeneous membranes would also be

observed.

It is clear that the presence of SS fillers improved the con-

ducting fraction of the membrane surface and consequently

improved the LCD and electrochemical behavior.

Desalination of NaCl by ED

The performance of composite membranes with 0.2 and 0.5 wt%

SS in a custom-designed ED cell was evaluated and compared

with the unmodified and commercial membranes. The ED test

J. Mater. Chem., 2011, 21, 7401–7409 | 7407

Fig. 7 i–v curve (a), and derivative dE/di as a function of i (b) and derivative dE/di as a function of E (c) of the obtained membranes measured at room

temperature in 0.025 mole dm�3.

Table 3 Characteristic values from i–v curve

Sample i/mA cm�2 i*/mA cm�2 DE/V

0SS 2.19 2.54 1.600.2SS 2.21 2.21 0.940.5SS 2.04 2.04 0.671SS 2.21 2.33 0.532SS 2.08 2.41 0.70FKE 2.06 2.08 1.10

Table 4 Performance of the prepared membranes compared withbenchmark in ED

Operating E/V SampleFlux/molem�2 h�1 h

P/kW hkg�1 salt

7 0SS 4.56 0.56 5.740.2SS 6.62 0.84 3.820.5SS 6.73 0.79 4.01FKE 7.46 0.83 3.92

5 0.5SS 4.12 0.99 2.28FKE 4.10 0.75 3.11

Publ

ishe

d on

11

Apr

il 20

11. D

ownl

oade

d by

Car

negi

e M

ello

n U

nive

rsity

on

22/1

0/20

14 1

8:07

:12.

View Article Online

was carried out with potentiostatic mode (CV) at 7 volts. The

concentration of desalting solution, pH and current density were

recorded over the testing period and the results were shown in

Fig. 8. The average values over the testing time of flux, current

efficiency and energy consumption were then calculated and

listed in Table 4. The composite membranes showed significantly

better performance compared to the pristine membrane, due to

the remarkable enhanced conductivity and transport properties

by incorporating surface functionalized silica. It is also inter-

esting to note that under the same testing condition, the

composite membranes also show comparable performance to the

benchmark membranes. Unlike the commercial membrane FKE,

which experienced the pH change after 60 min of operation, our

membranes were quite stable and the pH change was observed

Fig. 8 Electrodialysis of NaCl solution: changes of (a) c

7408 | J. Mater. Chem., 2011, 21, 7401–7409

for 0.5 wt% SS after 240 min. To ensure that the pH change was

not because of the chemical degradation of the membranes

during the ED operation, the IR spectra of the membranes

before and after ED testing were investigated (see ESI, Fig. S2†).

It was clear that there were no apparent difference of the

chemical structure observed. This implies that water dissociation

was occurring with the commercial and 0.5SS membranes, an

inappropriate testing condition for these two membranes.

Therefore, the lower potential condition was used for FKE and

0.5SS membrane to avoid water splitting phenomena. The results

are also shown in Table 4. In the later situation, the 0.5SS

membrane performed notably better than the commercial one

with 0.99 current efficiency and 2.28 kW h kg�1 of salts removed.

oncentration, (b) pH and (c) current density vs. time.

This journal is ª The Royal Society of Chemistry 2011

Publ

ishe

d on

11

Apr

il 20

11. D

ownl

oade

d by

Car

negi

e M

ello

n U

nive

rsity

on

22/1

0/20

14 1

8:07

:12.

View Article Online

As the performance of ED not only depends on membrane

properties but also the testing conditions and cell design, the

optimal operating condition of a given testing cell should be

carefully tuned.

Conclusion

A new two-step phase inversion membrane formation technique

allows a good control of the membrane structure, porosity and

electrochemical properties. Such properties were found to be

significantly affected by the presence of inorganic fillers. The

composite membranes with 0.2–0.5 wt% of SS showed the

optimum properties and performance in ED, which is compa-

rable to the commercial benchmark membranes. Incorporating

sulfonated mesoporous silica was shown to be an effective

approach to enhance the performance of ion-exchange

membranes. The findings from this work will lead to a better

understanding of the effects of inorganic fillers on sPES

membranes and provide new insights into how to develop

composite ion-exchange membranes with desirable properties for

desalination.

Acknowledgements

The financial support from Australian Research Council

(through its ARC Centre of Excellence and Discovery

programs), CSIRO Advanced Membranes for Water Treatment

Cluster Project, and Thai government (through Higher Educa-

tional Strategic Scholarship for Frontier Research Network) is

gratefully acknowledged. Additional thanks are due to Richard

and Robyn Webb for technical support on the SEM sample

preparation and TEM.

References

1 G. S. Gohil, R. K. Nagarale, V. V. Binsu and V. K. Shahi, J. ColloidInterface Sci., 2006, 298, 845–853.

2 J. Balster, O. Krupenko, I. Punt, D. F. Stamatialis and M. Wessling,J. Membr. Sci., 2005, 263, 137–145.

3 R. Scherer, A. M. Bernardes, M. M. C. Forte, J. Z. Ferreira andC. A. Ferreira, Mater. Chem. Phys., 2001, 71, 131–136.

4 M.-S. Kang, Y.-J. Choi, I.-J. Choi, T.-H. Yoon and S.-H. Moon, J.Membr. Sci., 2003, 216, 39–53.

5 A. Elattar, A. Elmidaoui, N. Pismenskaia, C. Gavach andG. Pourcelly, J. Membr. Sci., 1998, 143, 249–261.

6 Y. S. Dzyazko and V. N. belyakov, Desalination, 2004, 162, 179–189.7 S. Feng, Y. Shang, X. Xie, Y. Wang and J. Xu, J. Membr. Sci., 2009,335, 13–20.

8 H. K. Hansen, L. M. Ottosen and A. Villumsen, Sep. Sci. Technol.,1999, 34, 2223–2233.

9 C. Iojoiu, M. Marechal, F. Chabert and J.-Y. Sanchez, Fuel Cells,2005, 5, 344–354.

10 N. N. Krishnan, H.-J. Kim, M. Prasanna, E. Cho, E.-M. Shin, S.-Y. Lee, I.-H. Oh, S.-A. Hong and T.-H. Lim, J. Power Sources,2006, 158, 1246–1250.

This journal is ª The Royal Society of Chemistry 2011

11 G. Chamoulaud and D. B�elanger, J. Colloid Interface Sci., 2005, 281,179–187.

12 R. Marschall, I. Bannat, J. Caro and M. Wark, MicroporousMesoporous Mater., 2007, 99, 190–196.

13 R. Marschall, I. Bannat, A. Feldhoff, L. Wang, G. Q. M. Lu andM. Wark, Small, 2009, 5, 854–859.

14 H. Dai, R. Guan, C. Li and J. Liu, Solid State Ionics, 2007, 178, 339–345.

15 R. K. Nagarale, G. S. Gohil, V. K. Shahi, G. S. Trivedi andR. Rangarajan, J. Colloid Interface Sci., 2004, 277, 162–171.

16 I. S. Byun, I. C. Kim and J. W. Seo, J. Appl. Polym. Sci., 2000, 76,787–798.

17 J. Choi, K. M. Lee, R. Wycisk, P. N. Pintauro and P. T. Matherc, J.Electrochem. Soc., 2010, 157, B154–B159.

18 V. K. Shahi, Solid State Ionics, 2007, 177, 3395–3404.19 R. K. Nagarale,W. Shin and P. K. Singh,Polym. Chem., 2009, 1, 388–

408.20 L. Li and Y. Wang, J. Power Sources, 2006, 162, 541–546.21 C. L. Buquet, K. Fatyeyeva, F. Poncin-Epaillard, P. Schaetzel,

E. Dargent, D. Langevin, Q. T. Nguyen and S. Marais, J. Membr.Sci., 2010, 351, 1–10.

22 G. Alberti, M. Casciola, M. Pica and G. D. Cesare, Ann. N. Y. Acad.Sci., 2003, 984, 208–225.

23 V. V. Binsu, R. K. Nagarale and V. K. Shahi, J. Mater. Chem., 2005,15, 4823–4831.

24 R.-Q. Fu, L. Hong and J. Y. Lee, Fuel Cells, 2007, 1, 52–61.25 C. Li, G. Sun, S. Ren, J. Liu, Q. Wang, Z. Xu, H. Sun and W. Jin,

J. Membr. Sci., 2006, 272, 50–57.26 K. Miyatake, T. Tombe, Y. Chikashige, H. Uchida and M. Watanbe,

Angew. Chem., Int. Ed., 2007, 46, 6646–6649.27 S. P. Nunes, B. Ruffmann, E. Rikowski, S. Vetter and K. Richau,

J. Membr. Sci., 2002, 203, 215–225.28 C. Klaysom, R. Marschall, B. P. Ladewig, G. Q. M. Lu and L. Wang,

J. Mater. Chem., 2010, 20, 4669–4674.29 R. K. Nagarale, G. S. Gohil and V. K. Shahi, Adv. Colloid Interface

Sci., 2006, 119, 97–130.30 J.-H. Choi, S.-H. Kim and S.-H.Moon, J. Colloid Interface Sci., 2001,

241, 120–126.31 C. Klaysom, S.-H. Moon, B. P. Ladewig, G. Q. M. Lu and L. Wang,

J. Membr. Sci., 2011, 371, 37–44.32 M. Wilhelm, M. Jeske, R. Marschall, W. L. Cavalcanti, P. Tolle,

C. Kohler, D. Koch, T. Frauenheim, G. Grathwohl, J. Caro andM. Wark, J. Membr. Sci., 2008, 316, 164–175.

33 L. Chazeau, C. Gauthier, G. Vigier and J. Y. Cavaille, inHandbook ofOrganic–Inorganic Hybrid Materials and Nanocomposites, ed. H. S.Nalwa, American Scientific Publishers, 2003, vol. 2,Nanocomposites, pp. 63–111.

34 C. Barth, M. C. Goncalves, A. T. N. Pires, J. Roeder and B. A. Wolf,J. Membr. Sci., 2000, 169, 287–299.

35 J. Barzin and B. Sadatnia, Polymer, 2007, 48, 1620–1631.36 A. Conesa, T. Gumi and C. Palet, J. Membr. Sci., 2007, 287, 29–40.37 P. Van de Witte, P. J. Dijkstra, J. W. A. Van de Berg and J. Feijen,

J. Membr. Sci., 1996, 117, 1–31.38 A. Noshay and L. M. Robeson, J. Appl. Polym. Sci., 1976, 20, 1885–

1903.39 J. Balster, M. H. Yildirim, D. F. Stamatialis, R. Ibanez,

R. G. H. Lammertink, V. Jordan and M. Wessling, J. Phys. Chem.,2007, 111, 2152–2165.

40 R. Ibanez, D. F. Stamatialis and M. Wessling, J. Membr. Sci., 2004,239, 119–128.

41 N. Pismenskaya, P. Sistat, P. Huguet, V. Nikonenko andG. Pourcelly, J. Membr. Sci., 2004, 228, 65–76.

42 V. M. Aguilella, S. Mafe, J. A. Manzanares and J. Pellicer, J. Membr.Sci., 1991, 61, 177–190.

J. Mater. Chem., 2011, 21, 7401–7409 | 7409